尋找硬式即時系統的最差案例執行時間是相當重要的。我們只能從最差案例執行時間的靜態分析的方式找到一個最差案例執行時間的上限以取得有保證性的最後期限，然而業界還是依賴測量分析的方法來做最差案例執行時間的分析，甚至是用在硬式即時系統的分析上，主要是因為目前靜態分析工具還不是發展得很成熟。
我們使用SWEET (SWEdish Execution Time tool)來提供在ARM架構上的最差案例執行時間分析的支援。SWEET是由Mälardalen即時系統研究中心所開發的靜態最差案例時間分析工具。我們修改ARM-GCC以產生給SWEET的特定格式的輸入檔案: ALF, TCD, MAP。除此之外為了提供最差案例執行時間取向的最佳化支援和解決過度最佳化問題，我們也修改SWEET以獲得最差案例流向路徑和第二差案例等等的額外資訊。
經由來自[1]的驗證檔案的測試，我們的修改過的ARM-GCC能夠產生正確的輸入檔案讓SWEET分析，我們的修改過的SWEET也能夠提供額外的最差案例資訊。

Finding the worst-case execution time (WCET) on a hard real-time system is extremely important. Only static WCET analysis can give us an upper bound of WCET which guarantees the deadline, however, industrial practice still relies on measurement-based WCET analysis, even for many hard real-time systems; because static analysis tools are not a mature technology.
We use SWEET (SWEdish Execution Time tool) to provide WCET analysis support for the ARM. SWEET is a static WCET analyzer developed by the Mälardalen Real-Time Research Center (MRTC). We modified ARM-GCC to obtain input files in specific format for SWEET: ALF, TCD, and MAP. Besides, for WCET optimization supporting and over-optimizing issue, we modified SWEET to obtain additional worst-case flow path (WCFP) and the second worst-case information.
By testing with benchmark files from [1], our modified ARM-GCC can create correct input files for SWEET, and also the modified SWEET can produce additional worst-case information.

論文審定書..................................................................................i
摘要............................................................................................iii
Abstract.....................................................................................iv
Index............................................................................................v

1. Introduction 1
1.1. Real-Time Systems 1
1.2. Static Analysis of Worst-Case Execution Time 4
1.3. Static Analysis Support for ARM9TDMI 6
1.4. Additional Information for WCET Optimization 8
2. Concept & Related Work 10
2.1. Flow Analysis 10
2.2. Low-Level Analysis 13
2.3. Estimate Calculation 14
2.4. SWEdish Execution Time Tool (SWEET) 16
2.5. GandALF Converter 19
2.6. Scratchpad Memory Allocator 19
3. Implementation of Converting to TCD 22
3.1. Basic-Block Information in TCD 25
3.2. TCD Instruction Format 27
4. Implementation of Producing MAP 34
5. Implementation of Converting to ALF 36
5.1. ALF Format 37
5.2. Information not in GIMPLE 46
6. Information for Worst-Case Optimization 51
6.1. Worst-Case Flow Path 51
6.2. The Second Worst-Case Information 53
7. Results 56
7.1. ALF Intermediate Code 56
7.2. TCD Object Code 58
7.3. The Second Worst-Case Information 58
8. Conclusion and Future Work 60
8.1. Correctness of WCET 60
8.2. SPM Supporting in Low-Level SWEET 60
9. References 62

[1] WCET Benchmarks, Website: http://www.mrtc.mdh.se/projects/wcet/benchmarks.html
[2] SWEET (SWEdish Execution Time tool), Website: http://www.mrtc.mdh.se/projects/wcet/sweet.html
[3] ARM Information Center, Website: http://infocenter.arm.com/index.html
[4] ARM Information Center, ARM Instruction Set Quick Reference Card
[5] Andreas Ermedahl, Jan Gustafsson, and Björn Lisper, “ALF (ARTIST2 Language for Flow Analysis) Specification”, Technical Report, MRTC, Sep. 2008
[6] Andreas Ermedahl and Jakob Engblom, “Execution Time Analysis for Embedded Real-Time Systems”, Handbook of Real-Time Embedded Systems, p 35.1 -- 35.17, Chapman & Hall/CRC - Taylor and Francis Group, ISBN: 1-58488-678-1, Editor(s): Insup Lee, Jospeh Y-T. Leung, Sang H. Son, 2007
[7] Björn Lisper, Andreas Ermedahl, Dietmar Schreiner (external), Jens Knoop (external), Peter Gliwa (external), “Practical Experiences of Applying Source-Level WCET Flow Analysis on Industrial Code”, Proc. 4th$ International Symposium on Leveraging Applications of Formal Methods (ISOLA'10), p 449-463, Springer-Verlag, Heraclion, Crete, Editor(s):Tiziana Margaria and Bernhard Steffen, Oct. 2010
[8] Cheng-Ying Wu, “A Stack-Optimized Scratchpad Memory Allocator for Reducing Either the Average-Case or the Worst-Case Execution Time”, Computer Science and Engineering, NSYSU, 2008
[9] Cheng-Yu Yen, “Worst Case Execution time Analysis Support for the ARM Processor Using GCC”, Computer Science and Engineering, NSYSU, 2009
[10] Jia-yu Bai, “A Memory-Realistic SPM Allocator with WCET/ACET Tunable Performance”, Computer Science and Engineering, NSYSU, 2009
[11] J. Merrill, “GENERIC and GIMPLE: a new tree representation for entire functions”, In GCC developers summit 2003 (2003), pp. 171-180.
[12] Jakob Engblom, “Processor Pipelines and Static Worst-Case Execution Time Analysis”, Acta Universitatis Upsaliensis. Uppsala dissertations from the Faculty of Science and Technology 36. 130 pp. Uppsala. ISBN 91-554-5228-0
[13] Jan Gustafsson, Andreas Ermedahl, Björn Lisper, Christer Sandberg, and Linus Källberg, “ALF–A LANGUAGE FOR WCET FLOW ANALYSIS”, School of Innovation, Design and Engineering, Mälardalen University, Box 883, S-721 23 Västerås, Sweden, Nov. 2009
[14] Richard Stallman, “GNU Compiler Collection (GCC) Internals”, Free Software Foundation, 2008
[15] Stefan Bygde, “Static WCET Analysis based on Abstract Interpretation and Counting of Elements”, Licentiate Thesis, Mälardalen University Press, Mar., 2010
[16] Stefan Bygde, Andreas Ermedahl, Björn Lisper, ” An Efficient Algorithm for Parametric WCET Calculation”, The 15th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications, RTCSA 2009, p 13-21, IEEE Computer Society, Beijing, China, Editor(s):Patrick Kellenberger, Aug. 2009
[17] Steve Furber, “ARM system-on-chip architecture”, Addison-Wesley, ISBN 0-201-67519-6
[18] Wilhelm, R., Engblom, J., Ermedahl, A., Holsti, N., Thesing, S., Whalley, D., Bernat, G., Ferdinand, C., Heckmann, R., Mitra, T., Mueller, F., Puaut, I., Puschner, P., Staschulat, J., and Stenstrぴom, P. 2008. “The worst-case execution-time problem—overview of methods and survey of tools”, ACM Trans. Embedd. Comput. Syst. 7, 3, Article 36 (April 2008), 53 pages. DOI = 10.1145/1347375.1347389 http://doi.acm.org/10.1145/1347375.1347389